News
  New! April 2024: New paper on arXiv: Disguised Copyright Infringement of Latent Diffusion Models.
  New! Feb 2024: One paper accepted to SaTML 2024 (presented on April 11th).
  New! Jan 2024: I am a winner of the Cheriton Scholarship.
  New! Dec 2023: I am selected as a top reviewer in NeurIPS 2023.
  Oct 2023:One paper accepted to NeurIPS 2023.
  Oct 2023:One paper accepted to TMLR.
  May 2023: One paper accepted to ICML 2023.
  Dec 2022: One paper accepted to TMLR.
  Feb 2022: I am joining Huawei Noah's Ark Lab in Montreal as an Intern.
  January 2021: I am joining National Research Council of Canada as an Intern
  August 2020: I am joining University of Waterloo and Vector Instituite as a Ph.D. student
|
Research
I'm interested in trustworthy machine learning, especially how external training data affect ML models' performance and robustness,
i.e., the influence of data poisoning attacks (in the form of malicious external data) and beyond (e.g., in the form of disguised copyrighted material).
|
|
Disguised Copyright Infringement of Latent Diffusion Models
Yiwei Lu,
Matthew Y.R. Yang,
Zuoqiu Liu,
Gautam Kamath,
Yaoliang Yu
arXiv
We reveal the threat of disguised copyright infringment of latent diffusion models,
where one constructs a disguise that looks drastically different from the copyrighted sample yet still induces the effect of training Latent Diffusion Models on it.
|
|
Indiscriminate Data Poisoning Attacks on Pre-trained Feature Extractors
Yiwei Lu,
Matthew Y.R. Yang,
Gautam Kamath,
Yaoliang Yu
IEEE SaTML 2024
arXiv
We study indiscriminate data poisoning attacks against pre-trained feature extractors for fine-tuning and transfer learning tasks
and propose feature targeted attacks to address optimization difficulty under constraints.
|
|
Understanding Neural Network Binarization with Forward and Backward Proximal Quantizers
Yiwei Lu,
Yaoliang Yu,
Xinlin Li,
Vahid Partovi Nia
NeurIPS 2023
paper
We propose forward backward proximal quantizers for understanding approximate gradients in neural network quantization and provide a
a new tool for designing new algorithms.
|
|
f-MICL: Understanding and Generalizing InfoNCE-based Contrastive Learning
Yiwei Lu*,
Guojun Zhang*,
Sun Sun,
Hongyu Guo,
Yaoliang Yu
Transactions on Machine Learning Research
paper
We propose a general and novel loss function on contrastive learning based on f-mutual information. Additionally, we propose a f-Gaussain similarity funcntion with better
interpretability and empirical performance.
|
|
CM-GAN: Stablizing GAN Training with Consistency Models
Haoye Lu,
Yiwei Lu,
Dihong Jiang,
Spencer Ryan Szabados,
Sun Sun,
Yaoliang Yu
ICML 2023 Wrokshop on Structured Probabilistic Inference & Generative Modeling
paper
We propose CM-GAN by combining the main strengths of diffusions and GANs while mitigating their major drawbacks.
|
|
Exploring the Limits of Model-Targeted Indiscriminate Data Poisoning Attacks
Yiwei Lu,
Gautam Kamath,
Yaoliang Yu
ICML , 2023
paper
We find (1) existing indiscriminate attacks are not well-designed (or optimized), and we reduce the performance gap with a new attack; (2) there exists some intrinsic barriers of data poisoning attacks, namely that when the poisoning fraction is smaller than a (easy to calculate) threshold, no attack succeeds.
|
|
Indiscriminate Data Poisoning Attacks on Neural Networks
Yiwei Lu,
Gautam Kamath,
Yaoliang Yu
Transactions on Machine Learning Research (also appeared in NeurIPS 2022 ML Safety Workshop and Trustworthy and Socially Responsible Machine Learning (TSRML) Workshop)
paper/
code
We find that neural networks are surprisingly hard to (indiscriminate) poison and give better attacks.
|
|
f-mutual Information Contrastive Learning
Guojun Zhang*,
Yiwei Lu*,
Sun Sun,
Hongyu Guo,
Yaoliang Yu
NeurIPS 2021 workshop on self-supervised learning   (Contributed Talk)
paper/
poster/
talk
We propose a general and novel loss function on contrastive learning based on f-mutual information.
|
|
AdaCrowd: Unlabeled Scene Adaptation for Crowd Counting
Mahesh Kumar Krishna Reddy,
Mrigank Rochan,
Yiwei Lu,
Yang Wang
IEEE Transactions on Multimedia (TMM), 2021  
arXiv /
code
We propose a new problem called unlabeled scene adaptive crowd counting.
|
|
Few-shot Scene-adaptive Anomaly Detection
Yiwei Lu,
Frank Yu,
Mahesh Kumar Krishna Reddy,
Yang Wang
ECCV, 2020   (Spotlight)
arXiv /
code
We propose a more realistic problem setting for anomaly detection in surveillance videos and solve it using a meta-learning based algorithm.
|
|
Structure Learning with Similarity Preserving
Zhao Kang,
Xiao Lu,
Yiwei Lu,
Chong Peng,
Wenyu Chen,
Zenglin Xu
Neural Networks, 2020
arXiv
We propose a structure learning framework that retains the pairwise similarities between the data points.
|
|
Future Frame Prediction Using Convolutional VRNN for Anomaly Detection
Yiwei Lu,
Mahesh Kumar Krishna Reddy,
Seyed shahabeddin Nabavi,
Yang Wang
AVSS, 2019
arXiv /
code
We propose a novel sequential generative model based on variational autoencoder (VAE) for future frame prediction with convolutional LSTM (ConvLSTM).
|
|
Similarity Learning via Kernel Preserving Embedding
Zhao Kang,
Yiwei Lu,
Yuanzhang Su,
Changsheng Li,
Zenglin Xu
AAAI, 2019
PDF
We propose a novel similarity learning framework by minimizing the reconstruction error of kernel matrices, rather than the reconstruction error of original data adopted by existing work.
|
|
Semantic Segmentation in Compressed Videos
Yiwei Lu,
Ang Li,
Yang Wang
MMSP, 2019
PDF
We propose a ConvLSTM-based model to perform semantic segmentation on compressed videos directly. This significantly speed up the training and test speed.
|
|